Condensed-Phase Ethanol Conversion to Higher Alcohols - Industrial

Subscriber access provided by EPFL | Scientific Information and Libraries

Article

Condensed Phase Ethanol Conversion to Higher Alcohols Tyler L. Jordison, Carl T. Lira, and Dennis J. Miller Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.5b02409 • Publication Date (Web): 09 Oct 2015 Downloaded from http://pubs.acs.org on October 11, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Condensed Phase Ethanol Conversion to Higher Alcohols Tyler L. Jordison, Carl T. Lira, and Dennis J. Miller1 Department of Chemical Engineering and Materials Science Michigan State University East Lansing, Michigan 48824 (Revision #1 Sept, 2015)

Abstract Higher alcohols (C4+) can be formed from ethanol via condensation pathways collectively known as Guerbet reactions. Most prior Guerbet reaction studies involve vapor phase reactions, with n-butanol yields typically no higher than 30% of theoretical. We report here condensed-phase Guerbet reactions of ethanol over Ni/γ-Al2O3 catalysts modified by La2O3. Higher alcohol selectivities in excess of 80% at 230oC and autogeneous pressures are obtained in batch autoclave reactions. At these conditions, which are near the critical temperature of ethanol, the liquid phase is significantly expanded, byproduct gases (CH4 and CO2) are significantly dissolved in the liquid phase, and the vapor phase contains significant quantities of alcohols. To accurately compute ethanol conversion and product yields, both composition and quantity of each phase present at reaction conditions must be determined. To do this, the S-R Polar equation of state is combined with chromatographic analysis of liquid phase samples taken during reaction to model the phase equilibrium in the reactor at reaction conditions. Composition, density, and total number of moles of the vapor and liquid phases in the reactor are determined from the model and analysis, and are used to calculate more accurate values of conversion and product yield than those calculated by liquid phase samples alone.

1

To whom correspondence should be addressed at [email protected], (517) 353-3928.

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 31

1. Introduction The continuing trend in U.S. policy toward production of renewable fuels has led to the maturation of corn ethanol production and the recent emergence of commercial cellulosic ethanol. In addition to being a biofuel, ethanol is a valuable chemical feed stock that could penetrate existing markets as its production increases. One such market is the production of butanol and higher alcohols.

These alcohols have broad applications in the commercial

chemical industry as feed stocks and components for solvents, consumer goods, and materials.1 As fuels, higher alcohols have energy values closer to gasoline than ethanol and have lower affinity for water, making them superior as fuel components to ethanol (Table 1).2 It is no surprise, therefore, that butanol and higher alcohols have higher market value than ethanol.3 Table 1 Lower heating values of higher alcohols. The solubilities of alcohols in water, and solubilities of water in the alcohols, are also shown.

LHV (MJ/L) 21.1 26.8 29.3 29.9 31.0 31.3 32.52

Ethanol 1-Butanol 1-Hexanol 2-Ethyl-1-Butanol 1-Octanol 2-Ethyl-1-Hexanol Gasoline

Solubility at 20°C (wt% solution) of H O in H O4 2

100 20.1 7.2 4.6 2.6

2

100 7.8 0.58 0.43 0.06 0.10

The current method for producing 1-butanol is the oxo process, in which petroleumderived propylene1 is hydroformylated (CO + H2) over a homogeneous rhodium catalyst to form butryaldehyde, which is then hydrogenated to produce 1-butanol. The oxo process is complicated, requires high energy input, and is costly,1 and butanol prices fluctuate with propylene prices.


Page 3 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The direct conversion of ethanol to butanol would be a renewable, environmentally “green” process that could become more attractive economically than the oxo process. Ethanol production technologies, whether utilizing sugar cane, starches, or cellulose, are constantly becoming more efficient and less expensive. Hence, ethanol prices will continue to decline over time, while oil (and propylene) prices will almost inevitably increase.

Figure 1 Reaction tree for the ethanol Guerbet reaction system.

Reactions involving carbon-carbon coupling of alcohols are commonly regarded as Guerbet reactions.5-8 Guerbet first reported the reaction of ethanol to butanol in the late 19th century, but achieved only a low yield.9 In the Guerbet reaction, ethanol is oxidized to



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

acetaldehyde; acetaldehyde undergoes an aldol condensation to crotonaldehyde, and then the crotonaldehyde is hydrogenated to butanol.10 One molecule of water is formed for every molecule of butanol formed. The formed butanol can then react with itself or with ethanol to form 1-hexanol, 2-ethyl-1-butanol, 2-ethyl-1-hexanol, and other higher alcohols. A Guerbet reaction tree is shown in Figure 1, which gives details of the aldol condensation mechanism for these higher alcohol reactions. A multifunctional catalyst is required for Guerbet reactions that has a metal component for hydrogenation/dehydrogenation and the proper balance of acidic and basic sites to facilitate aldol condensation chemistry. It is noted that some literature supports this aldol condensation (Guerbet) mechanism,1,1113

while other authors support a simpler proton abstraction mechanism.10,14,15 In proton

abstraction, a proton is extracted from the beta-carbon of one ethanol molecule, which produces a nucleophile that attacks another ethanol molecule.10 The coupled product dehydrates to form 1-butanol and water. Many researchers have converted ethanol to butanol in the vapor phase at relatively high temperatures (>300°C), achieving ethanol conversions of 7-80% and 1-butanol selectivities of 10-70%.1,8,10,12,15-28 Others have used ethanol as a limiting reagent with methanol and/or propanol to attain moderate to high ethanol conversion.3,7,22,29,30 Catalyst composition is the most important factor in higher alcohol yields - many studies used MgO in pure form or mixed with other basic oxides as a solid base catalyst for this reaction.7,8,10,16,21,22,25,30 Alkali cationexchanged zeolites have also been used.14 The most recent literature describes partially decomposed hydrotalcites, hydroxyapatites, and γ-Al2O3 -supported nickel, which have produced the highest butanol yields thus far.11,15,18


Page 4 of 31

Page 5 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Few researchers have performed ethanol Guerbet reactions in the condensed phase. Riitonen et al. looked at γ-Al2O3-supported metal catalysts in a batch reactor at 250°C and pressures up to 100 bar.31 Ethanol conversion and 1-butanol selectivity reached maximum values of 25% and 80%, respectively. The same authors then carried out continuous liquid-phase ethanol conversion to 1-butanol at 240°C and 70 bar.26 With continuous operation, 1-butanol selectivity was 70% with a commercial Ni/γ-Al2O3 catalyst and ethanol conversion was between 10 and 30%. Cobalt supported on γ-Al2O3 produced the highest ethanol conversion. Ghaziaskar et al achieved 35% ethanol conversion and 83% C4+ alcohol selectivity at 250°C and 176 bar with 8% Ni/γ-Al2O3.17 Although heterogeneous catalysts have been most widely studied, Wass et al. claimed 95% selectivity to 1-butanol with a homogeneous ruthenium diphosphine catalyst.32 Mixed oxides, such as MgxAlOy, contain weak Lewis acid-strong Bronstead base site pairs necessary for ethanol dehydrogenation, aldol condensation, and butyraldehyde hydrogenation.6,33 This functionality can be replicated by a more stable and active γ-Al2O3supported nickel catalyst. This was first demonstrated by Yang et al., who used an 8wt% Ni/γAl2O3 to achieve a vapor phase ethanol conversion of 19% with 64% selectivity to 1-butanol.15 The nickel metal provides dehydrogenation/hydrogenation functionality, while the acid/base sites on the γ-Al2O3 work in conjunction with the nickel to perform the aldol condensation step in the Guerbet mechanism.6,33 The high selectivity achieved with nickel on γ-Al2O3 catalyst makes it a good starting point for this work, which investigates modification of the nickel catalyst with lanthanum oxide for condensed-phase ethanol Guerbet reactions. It is hypothesized that lanthanum will inhibit side reactions by adding basicity to the γ-Al2O3. Liquid phase processing is intrinsically



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

preferable to the vapor phase route, as it involves smaller reaction vessels for the same throughput and saves energy by not requiring feed vaporization. Because the reaction conditions for liquid-phase Guerbet reactions are close to the critical temperature (241°C) and critical pressure (6.3 MPa) of ethanol,34 the liquid phase is significantly expanded and non-ideal. The vapor phase is also dense, and thus significant quantities of alcohols are present there. Proper characterization of the reaction thus requires that both phases be included in yield and conversion calculations. Therefore, the SchwartzentruberRenon-Polar equation of state35 has been implemented to predict higher alcohol phase equilibrium at these reaction conditions. Equations of state (EOS) account for the volume of the molecules as well as the interactions between molecules,36 and are more suited than activity models for predicting P-V-T behavior of pure components and mixtures at near-critical to supercritical conditions.

2. Experimental 2.1 Materials and Catalyst Preparation Ni(NO3)2·6H2O (Reagent Grade, Jade Scientific), and La(NO3)2·6H2O (99%, Fluka) were used as catalyst precursors. The supports used include 1/8” x 1/8” γ-Al2O3 cylindrical extrudates (Johnson Matthey) and 1/16” diameter γ-Al2O3 spheres (Strem Chemical). Anhydrous ethanol (Koptec, 200 proof) was used as the initial reactor charge. The catalysts prepared (identifier) were 8 wt% Ni/γ-Al2O3 (8Ni/Al), 10 wt% Ni/γ-Al2O3 (10Ni/Al), 8 wt% Ni/7 wt% La2O3- γAl2O3 (8Ni/7La-Al), 14 wt% La2O3- γ-Al2O3 (14La-Al), 8 wt% Ni/9 wt% La2O3- γ-Al2O3 (8Ni/9La-Al), and 8 wt% Ni/ 10 wt% CeO2 - γ-Al2O3 (8Ni/10Ce-Al). Nickel catalysts supported on γ-Al2O3 were prepared by incipient wetness impregnation of γ-Al2O3 using a pre-determined quantity of solution to fill the γ-Al2O3 pore volume. The catalysts (30 g per batch) were dried at 130°C for 18 hours and then reduced at 525°C and 1 atm 6 ACS Paragon Plus Environment

Page 6 of 31

Page 7 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


in a tubular flow reactor for 20 hours in 35 ml (STP) H2/min. Nickel catalysts modified with La2O3 were prepared in the same fashion as above, except that La(NO3)3 was deposited first by incipient wetness impregnation of the γ-Al2O3 support followed by drying at 130°C for 18 hours, and calcining at 600°C for 20 hours in 35 ml/min N2 flow. This assured there was La2O3 on the γ-Al2O3 surface before the impregnation of the nickel. Most of the catalysts for screening studies were prepared on 1/8” extrudates, which were crushed into smaller particles (0.3-0.8 mm) before use. Several later catalysts were prepared on 1/16” γ-Al2O3 spheres and used without crushing in batch reactions. 2.2 Catalyst Characterization Acid and base site densities on prepared catalysts were measured with a Micromeritics Autochem II chemisorption analyzer. Ammonia and CO2 were used for acid and base site adsorption, respectively. After loading and establishing a stable baseline, catalysts were degassed by ramping the temperature to 600°C at 10°C/min under helium and holding at 600°C for 60 minutes. Outgassed samples were cooled to 25°C at 10°C/min and held for 10 minutes. Ammonia or CO2 was then passed across the catalyst at a flow of 50 ml/min for 30 minutes. Gas flow was then changed to helium (50 ml/min) for 90 minutes to remove weakly bound gases. Desorption of CO2 or ammonia was carried out by ramping temperature to 600°C at 10 °C/min and holding at 600°C for 30 minutes. BET surface area measurements were done by nitrogen adsorption at 78K with Micromeritics ASAP 2010. Before analysis, samples were degassed in the degas port of the instrument at 120°C for 24 hr. Nickel dispersion was measured by H2 chemisorption at 35oC in the same instrument. For chemisorption, samples were heated under vacuum to 450oC, reduced in 1.0 atm H2 at 450oC for 2 hr, evacuated for 0.5 hr at 450oC, cooled under vacuum to 35oC, and then exposed to H2 at 35oC. 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2.3 Batch Reaction Studies Ethanol Guerbet reactions were performed in a 300ml Parr reactor (Model 4842, Parr Instruments, Chicago, Illinois) with reaction times between two and ten hours. Typically, 110 g of pure ethanol (or the desired feed mixture) was placed into the reactor along with the desired amount of catalyst. The reactor was purged with nitrogen and sealed with 0.1MPa of nitrogen overpressure. The reactions were carried out at autogeneous pressure. The Parr reactor uses an Omega 1/8” stainless steel Type J thermocouple which was calibrated with the boiling points at 745 mm Hg absolute pressure of water (99.45oC), 1,2 propylene glycol (187.0oC), and ethyl nonanoate (226oC).37 The controller measured a temperature of 98°C for water, 185°C for 1,2-propylene glycol, and 225oC for ethyl nonanoate. Based on these results, a correction of +1oC was applied to the temperature measured during experiment. This correction was included when comparing experimental results with results obtained from phase equilibrium modeling using the SR-Polar equation of state. Pressure measurements for all reactions used a pressure gauge on the reactor head. The pressure gauge had increments of 50 psi for pressures up to 3000 psig. The accuracy of this gauge was checked with a large 10 psi increment dial test gauge. To measure pressures in the range of reaction conditions, the reactor was evacuated to 14 torr and then filled with ethanol. Ethanol vapor pressure was measured from 190°C to 230°C, accounting for the reactor temperature correction mentioned above. The dial test gauge consistently read 1.5 ± 0.5% (5-10 ± 2-3 psi) above the ethanol vapor pressure calculated from Antoine’s equation37 and the reactor head gauge consistently read 2.8 ± 1.5% (10-20 ± 10 psi) below the calculated ethanol vapor pressure. A check of the effect of this error on product yields and selectivities from experiment showed less than a one percentage point difference in selectivity in the worst case, so no correction to measured pressure was invoked in analysis of the experiments. 8 ACS Paragon Plus Environment

Page 8 of 31

Page 9 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Initial catalyst screening experiments were performed at a catalyst loading of 0.093 g catalyst/g ethanol, a 280 rpm stir rate, and 230°C reaction temperature with only the final reaction mixture analyzed. In all later experiments, however, concentration profiles of key species over time were established by withdrawing liquid samples periodically from the reaction phase via a dip tube into an evacuated 1/8” x 8” stainless steel sample tube with a valve at each end to isolate the liquid sample from the reaction vessel. The sample tube was vented after isolating and the liquid sample was subjected to analysis by gas chromatography. Reactor pressure was monitored during reaction and after cooling at the end of reaction to aid in determining product compositions and quantities of gas formed. The quantity of gas produced in reaction was determined by weighing the entire cooled reactor with contents both before and after depressurization; the gas exhausted during depressurization was collected in a gas bag and analyzed by gas chromatography.

2.4 Analytical Liquid phase reaction samples were diluted 10-fold in acetonitrile and then analyzed on a Varian 450 gas chromatograph configured with a DBWax column (30 m x 0.53 mm ID, 1.0 µm film) and TCD detection. The temperature program for GC separation of liquid samples was to hold at 40°C for 2 min, ramp to 150°C at 10 °C/min, ramp to 250°C at 30°C/min, and then hold at 250°C for 2.00 min. Chromatographic response factors were determined by injecting calibration samples; in later experiments butyl hexanoate was used as an internal standard to improve analytical accuracy. A sample chromatogram is shown in Figure 2. Gas phase samples collected during depressurization at the end of experiment were analyzed on a Varian 3300 gas chromatograph with 60/80 Carboxen-1000 column (15 ft x 1/8” SS, 2.1mm ID) and argon carrier gas. The temperature program for GC separation of gas 9 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

samples was to hold at 40°C for 2.0 min, ramp to 250°C at 20 °C/min, and hold at 250°C for 5.0 min. A calibration gas mixture containing 2.0 vol% each of CO, CH4, and CO2, along with 100% CO2 and 100% CH4, were used to develop response factors for the gas analysis.

Figure 2 Sample chromatogram of liquid sample at end of run. All alcohol peaks and water peak are resolved.

Chromatographic data were entered into an Excel spreadsheet where calculations of species mass fractions in each liquid sample were carried out. This information was then incorporated into the thermodynamic modeling of the reaction system as described below, ultimately leading to determination of ethanol conversion and product selectivities.

3. Thermodynamic Modeling 3.1 SR Polar Equation of State Initial screening of several equations of state (see Supporting Information) led to the selection of the Schwartzentruber-Renon (SR)-Polar equation of state (EOS) to characterize vapor and liquid phases during batch reactor operation. Accurate modeling of phase equilibrium and liquid and vapor phase densities is important in the current application, but is generally challenging in the near-critical region.36 Molecular level simulations, in which macroscopic 10 ACS Paragon Plus Environment

Page 10 of 31

Page 11 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


properties are predicted a priori from molecular structure, consume large computing resources and are not yet able to predict experimental behavior. The SR-Polar EOS was thus chosen because it offers the advantage of a temperature dependent molar volume translation parameter to best predict densities, and is recommended for non-ideal systems at high temperatures and pressures. The SR Polar EOS was developed by Schwartzentruber and Renon in 1989.35 It is a Soave-Redlich-Kwong (SRK) type equation of state. The SR-Polar EOS, including volume translation, is =

− + − ( + )( + )

(1)

where a is the attractive parameter, b is a repulsive parameter, and c is the volume translation parameter. The species included in the SR-Polar EOS model of the reaction system are ethanol, 1-butanol, 1-hexanol, water, CH4, and CO2, with the ratio of CH4:CO2 determined by GC analysis of vented gas at the end of reaction. In reality, the reaction liquid phase contains minor amounts of other species including acetaldehyde, ethyl acetate, diethyl ether, and longer chain alcohols, aldehydes, and esters. For phase equilibrium modeling purposes, these components are combined with one of the species mentioned above according to volatility. The groupings are given in Table 2 below; unidentified peaks in sample chromatograms were assigned the response factor for 1-hexanol for calculation purposes, and any unaccounted carbon was assigned as 1hexanol for modeling and then subsequently removed to calculate yields based on the model results. Once compositions of the modeled components were established, their mole fractions were recalculated according to experimental results to give the modeled quantities of all species observed in Table 2.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table 2

Page 12 of 31

Grouping of observed species into modeled components. Component Modeled Component Ethanol Ethanol 1-Butanol 1-Butanol 1-Hexanol 1-Hexanol H2O H2O 2-Ethyl-1-Butanol 1-Hexanol 1-Octanol 1-Hexanol 2-Ethyl-1-Hexanol 1-Hexanol Diethyl Ether Ethanol Ethyl Acetate Ethanol Acetaldehyde Ethanol Butyraldehyde Ethanol

3.2 Parameter Estimation The SR-Polar EOS utilizes quadratic mixing rules for the attractive parameter a and the repulsive parameter b:

= ( )/ [1 − , − − ]

=

where

+ (1 − !, ) 2

" , = , + , + , / " = + + /

(2)

(3)

(4) (5)

For all binary pairs, kb,ij was set to zero and lij was set to zero except for the water/1butanol binary pair. The binary parameter in Eq. (4) was chosen to be constant, (ka,ij=ka,ij0) for ethanol/CO2, and 1-butanol/CO2. The binary pairs ethanol/1-butanol, and ethanol/water used ka,ij=ka,ij0+ ka,ij2/T. The water/1-butanol binary pair used ka,ij=ka,ij0+ ka,ij2/T and lij=lij0+ lij2/T. Ethanol/CH4 and water/CO2 were the only binary pairs that used a linear temperature dependence of ka,ij. 12 ACS Paragon Plus Environment

Page 13 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Values of the ka,ij and lij parameters were regressed from binary phase equilibrium experimental data in the AspenPlus database; parameters for the ethanol/CH4 binary pair only were later adjusted to better describe the experimental results. Values of the regressed parameters are summarized in Table S2 of Supporting Information. For binary pairs where experimental data were not available, parameters were estimated using UNIFAC or HaydenO’Connell models. The pure component parameters for the SR-Polar EOS are defined by:

=

#($/% &)

('()* )+ ,* -)*

2/. − 1 0

= 3 0

(6)

(7)

where Tci and Pci are critical properties of each species. The value of αi at subcritical temperature (Tri = T/Tci < 1) is described by the extended Mathias equation (Eq. 8); in the supercritical region the equation to determine αi is known as the Boston-Mathias extrapolation (Eq. 9).38 For alcohols and water, Tri is less than 1 and αi is defined as: $ +

1 = [1 + 2 31 − 4 5 − 6, (1 − 4 )1 + 6, 4 + 6., 4 ] when Tri

2 − 6, 1 + 6, + 6., 2 2 = 0.48508 + 1.55171E − 0.15613E > = 1 +


(9)

( 10 ) ( 11 ) ( 12 )


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 31

The parameter ωi is the acentric factor. For the polar components present in the ethanol Guerbet system, p1,i, p2,i, and p3,i are estimated by the Antoine equation.38 The SR-Polar EOS offers a temperature-dependent volume translation parameter c for the purpose of accurately predicting densities with a linear mixing rule:

=

( 13 )

Volume translation does not affect VLE composition calculations but does affect fugacity values.38 For alcohols and water, Tri < 1; therefore the pure component volume translation parameter ci for these components is defined by:

= " +

0$*

&(/()* GH+*

(for Tr 1, and c0,i was regressed from binary ethanol/CH4 and ethanol/CO2 liquid densities at conditions close to those of reaction (with c1,i and c2,i set to zero). Values of all regressed volume translation parameters are given in Table S3 of Supporting Information. 3.3 Applying SR-Polar EOS to Batch Experiments The SR-Polar EOS is applied to batch Guerbet chemistry experiments to determine compositions and quantities of both vapor and liquid phases. This is done by interfacing the AspenPlus V8.2 process simulator with Microsoft Excel 2013 via the AspenPlus Properties Addin in Excel. AspenPlus flash operations using TVFlash, particularly to identify bubble and dew point pressures at reaction temperature, are performed in Excel as described below after the Properties Backup (.aprbkp) file with SR-Polar binary constants and parameters is loaded into the Excel workbook. 14 ACS Paragon Plus Environment

Page 15 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


When a liquid phase sample taken during reaction is vented to atmospheric pressure, dissolved gases flash out; consequently, the GC analysis of the remaining material reflects only the composition of the condensable portion of the liquid phase at reaction conditions. Starting with this composition, a mixture of CO2 + CH4 with a fixed molar ratio (xCH4/xCO2) is added to the GC-determined liquid composition using AspenPlus TVFlash (maintaining Σxi = 1) until the bubble pressure of the combined mixture equals the measured reactor pressure at the point the sample was taken. This identification of bubble pressure defines the compositions, molecular weights, and densities of liquid and vapor phases at reaction conditions. Next, because the total mass of the reaction mixture (mT) is tracked during reaction, and the total mixture volume is constrained by the reactor volume (VR), the number of moles of vapor nV (and consequently moles of liquid nL) in the reactor can be determined by combining Eq. (15) and (16) to give Eq. (17), from which nV can be directly calculated and nL can be subsequently determined from Eq. (15) or Eq. (16).

2 ( = IJ KJ + IL KL IJ IL + MJ ML

( 16 )

' MJ KJ ML − 2 ( ML MJ KJ − ML KL

( 17 )

' =

IL =

( 15 )

Once vapor and liquid quantities and compositions are defined, the overall composition (zi) of the reaction mixture is calculated at the time of sampling:

P =

IJ + IL Q IRSR 15

ACS Paragon Plus Environment

( 18 )


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 31

where ntot is the sum of nV and nL. Once this has been done, the ratio of liquid-phase CH4 to CO2 mole fractions (xCH4/xCO2) chosen for the TVFlash calculation is then varied and the entire calculation is repeated until the calculated overall gas composition ratio (zCH4/zCO2) matches the overall experimental (zCH4/zCO2) ratio determined by GC analysis. The calculation of ethanol conversion and product selectivities and yields are then finally calculated by the following equations: `\VU^ a Yb ]c\deWX ^]UWYU^ Y

TUVUWXYZYX[ X\ T]UWYU^ Y (%) = ( `\VU^ a Yb cUfWXUd UXgfb\V ) x 100% `\VU^ \j UXgfb\V Yb cUfWX\c

hXgfb\V a\bZUc^Y\b % = (i − `\VU^ \j YbYXYfV UXgfb\V WgfckUd) x 100% lYUVd \j T]UWYU^ Y (%) = (

TUVUWXYZYX[ % )(a\bZUcY^\b %) imm%

( 19 )

( 20 ) ( 21 )

3.4 Model Validation and Parameter Adjustment The phase equilibrium model using the SR-Polar EOS for the batch reactor was validated by conducting non-reactive control experiments in which known quantities of the modeled components were placed into the Parr reactor without catalyst and the reactor pressure was measured at reaction temperature. Two multicomponent control experiments were conducted at 230oC: one simulating high ethanol conversion (49%) and the other low ethanol conversion (22%). For these experiments and all subsequent applications of the SR-Polar EOS, a volume VR = 328 ml for the Parr batch reactor was used. In the multicomponent control experiments, a defined liquid mixture of ethanol, 1butanol, 1-hexanol, and water was first added to the reactor, which was sealed and then weighed. The reactor atmosphere was purged with CO2 and pressurized to give the desired amount of CO2, and then weighed to verify the mass of CO2 added. The gas inlet was then purged with CH4, 16 ACS Paragon Plus Environment

Page 17 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


connected to the reactor, and pressurized to a pressure higher than the reactor pressure to prevent backfilling. Methane was then added by monitoring pressure change, and then the reactor was finally weighed again to determine the mass of CH4 added. The reactor was heated to reaction conditions, reactor pressure was recorded, and liquid samples were taken to determine composition. The SR-Polar EOS was then applied as described in the above section to calculate quantity, composition, and density of liquid and vapor phases in the reactor. The total quantities of materials present in the reactor were then calculated from Eq. (15) – (18) for comparison with the experimental quantities added. Initial examination of the results showed that the SR-polar EOS model underpredicted total gas quantity when using the binary parameters regressed from experimental data in the AspenPlus database. The binary VLE parameters for ethanol/CH4 and ethanol/CO2 were therefore checked by performing control experiments with just ethanol-CH4 and ethanol-CO2 binary mixtures in the batch reactor. Known quantities of the binary mixtures were heated and pressure was recorded at 215°C and 225°C. The bubble pressure calculation was performed using TVFlash in AspenPlus for each of the binary experiments by adjusting ka,EtOH-CH40 and ka,EtOH-CO20 until the difference between calculated and experimentally added quantity of gas was minimized. (At temperatures higher than 225°C, the binary bubble pressure calculation in AspenPlus did not converge because of the proximity to the ethanol critical point.) The best agreement between calculated and experimentally added quantities of CH4 in the reactor was obtained by adjusting the ethanol/CH4 binary interaction parameter from ka,EtOH-CH4 = 0.0030T1.1945 to ka,EtOH-CH4 = 0.0047T-2.3619 (T in K). Similarly, the best agreement with the ethanol/CO2 binary experiment was obtained by adjusting the binary parameter ka,EtOH-CO2,0 from +0.1 to -0.1.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 31

These refitted values of binary parameters were then used in the analysis of the multicomponent control experiments. With the refitted parameters, the difference between predicted and experimental CO2 + CH4 mass in the control experiments decreasing from 0.9 g to 0.6 g for the high conversion experiment, while the difference between predicted and actual total CO2 + CH4 mass in the reactor for the low conversion experiment decreased from 0.4 g to 0.2 g (Table 3). With this correction, calculated overall reactor composition and simulated ethanol conversion and higher alcohol selectivities were within three percentage points of the values based on known compositions of the control experiments, as seen in Table 3. The refitted values of ka,EtOH-CH4 and ka,EtOH-CO2 were therefore used in all subsequent applications of the SR-Polar EOS to reaction studies. Table 3 Comparison of predicted and experimental gas quantities in SR-Polar control experiments at 230°C with adjusted ethanol/CH4 and ethanol/CO2 binary parameters.

Component Ethanol 1-Butanol 1-Hexanol H2O

Overall mole fraction (zi) High Conversion Low Conversion Experimental Predicted Experimental Predicted 0.465 0.469 0.770 0.788 0.098 0.097 0.077 0.067 0.045 0.041 0.013 0.010 0.202 0.215 0.101 0.099

CH4

0.133

0.126

0.023

0.022

CO2

0.056

0.053

0.016

0.014

Total Mass of Species (g)

107.0

107.0

84.2

84.2

Observed Pressure (psia)

1565

1565

800

800

Liquid Species Mass (g)

95.1

95.7

82.2

82.4

CH4 + CO2 Mass (g)

11.9

11.3

2.0

1.8

Ethanol Conversion

0.485

0.476

0.219

0.194


Page 19 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4. Results and Discussion 4.1 Catalyst Characterization Catalysts used in reaction studies were subject to measurement of BET surface area via N2 adsorption, Ni metal site density via H2 chemisorption, acid site density via NH3 chemisorption, and basic site density via CO2 chemisorption.

Results are presented in Table 4; profiles from

temperature programmed desorption (TPD) experiments are given in the Supporting Information for both NH3 and CO2 (Figure S5). NH3 desorption from γ-Al2O3 characterizes weak (35-150°C), medium (150-300°C), and strong (300-600°C) acid sites, and CO2 desorption from γ-Al2O3 gives information on weak (35-150°C), medium (150-300°C), and strong (300-600°C) basic sites. These temperature ranges for relative base and acid site strengths have been assigned in prior work.13 Temperature programmed desorption results in Table 4 show that total acid site density is relatively constant in the several catalysts examined, although the distribution of acid site strengths shifts toward weaker acidity with addition of La2O3 to γ-Al2O3.

Basic site

concentration measured Table 4. Acidic, basic, and Ni metal site densities and total surface area of catalysts

Catalyst Acidic Sites (µmol g-1) Weak Medium Strong Total Basic Sites (µmol g-1) Weak Medium Strong Total Ni Metal Sites (µmol g-1) BET Surface Area (m2 g-1)

γ-Al2O3

8Ni/Al

9La-Al

8Ni/9La-Al

190 240 170 600

170 230 280 680

180 300 110 590

190 260 120 570

61 32 2 95

84 51 7 142 17 152

120 150 50 320

100 130 60 290 14 124

153


145


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 31

by CO2 adsorption, in contrast, shows the expected strong increase upon addition of La2O3, with a concurrent shift from weaker (35-150°C) to stronger (> 150°C ) basic sites at higher temperatures. 4.2

Calculation of Reacting Phase Volume

The SR-polar EOS describes reaction phase properties that cannot be observed or that are normally accounted for in batch reaction studies. Most reactions in this work were carried out at 230oC, close to the critical temperature of ethanol (241oC). Under these conditions, the liquid phase is significantly expanded and the vapor phase density is high enough that a significant fraction of the reaction mixture is present in the vapor phase. Figure 3 shows that at 230°C, pure ethanol has a liquid phase density of 0.44 g/ml, slightly more than half its value at 25oC, and the reaction mixture at partial conversion (right-most picture in Figure 3) is similarly expanded. This liquid phase expansion has kinetic implications in calculating species concentrations at reaction conditions, and in recognizing that vigorous stirring is required to suspend the solid catalyst in the low density liquid solution at reaction conditions. The expanded liquid reaction phase also requires activity-based kinetic models, because of its highly non-ideal nature. More importantly, the liquid expansion has safety implications, in that if the reactor is initially filled with too much ethanol, the liquid phase will expand and fill the reactor head space as reaction temperature is approached, leading to reactor overpressure and possibly reactor failure.


Page 21 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4.3 Catalyst Evaluation for Guerbet Reactions

Figure 3 Liquid and vapor phase densities for pure ethanol at 25°C (left), pure ethanol at 230°C (center), and for the reaction mixture at 41% ethanol conversion (right).

Analysis of reaction equilibria shows that all reactions are favorable except for dehydrogenation of ethanol to acetaldehyde, for which the equilibrium constant at reaction conditions is ~0.1 (see Supporting Information). The SR-Polar equation of state model has been applied to all reactions carried out and reported in this study. An early experiment carried out at 230oC with 8Ni/Al catalyst (0.09 g catalyst/g EtOH) is analyzed here (Table 5) to illustrate application of the model. In this run, samples were taken at ten time points. Ethanol conversion, product selectivities, and carbon recoveries were calculated after application of the SR-Polar equation to each time point. The carbon recovery was greater than 95% at every time point. The 1-butanol selectivity starts low at 6%, but increases to a maximum of 51%. The selectivity to CH4 and CO2 is high early in the reaction (

Condensed-Phase Ethanol Conversion to Higher Alcohols - Industrial

Recommend Documents